CN110517084B - Vehicle function activity analysis method and system - Google Patents

Abstract

The invention aims to provide an analysis method and system for the activity of an automobile function, which comprise the steps of original data acquisition, data processing, data analysis and data visualization, wherein the existing data acquisition hardware is utilized, a terminal monitors CAN bus data under the condition of not increasing cost and manpower, a whole automobile function signal is uploaded, data is received and stored in a background, and the use frequency and the use time of the automobile function are obtained through data operation, so that the activity of the function is obtained. The invention has the data real-time performance and the whole sample quantity, can reflect the functional activity of the actual user, forms the analysis of multiple dimensions, and meets the requirements of products and operators.

Description

Vehicle function activity analysis method and system

Technical Field

The invention relates to a vehicle function activity analysis technology.

Background

In recent years, the importance of data has become increasingly prominent in the process of converting a whole car factory from a conventional car-making enterprise to a service-type enterprise. The more and more functions the vehicle is mounted on, the more and more complicated functions the user may not be suitable for, and the whole vehicle factory has also invested a lot of research and development cost in order to carry these bright spot functions. However, how often the end user evaluates these functions, whether they are used as often as expected, and whether the user really needs them, is worth the intensive research to promote the functional design of the automobile. At present, a whole vehicle factory mostly adopts an off-line investigation mode to acquire evaluation data, and the mode has the defects of small sample size, poor universality, high cost and the like.

The data analysis theory of the existing terminal APP is very mature, products and operators can analyze the active condition of functions through data embedded points, find favorite functions of users, mine reasons behind the favorite functions of the users and whether potential relevance exists among the functions, and guide product design and research and development.

At present, a whole vehicle factory faces a whole vehicle CAN bus, a method for analyzing the activity of the whole vehicle function based on huge data of the CAN bus is developed, the behavior data analysis of an actual user is utilized to replace the traditional user investigation, the cost is reduced, the sample size is greatly enlarged, and the authenticity of the data is ensured.

However, in the existing method for analyzing the activity of the whole vehicle function based on huge data of the CAN bus, the data is transmitted in an off-line mode of local storage and off-line copying, the data does not have real-time performance, and the sample size is still limited by the storage and copying capacity.

Therefore, how to grasp the requirements and the active conditions of the users on the functions of the vehicle in real time and dynamically provides the users with services and functions closer to the requirements, and finds out the bright spots and the vinasse spots from a plurality of functions for the whole vehicle factory, thereby becoming the problem that the whole vehicle factory needs to face in development.

Disclosure of Invention

The invention aims to provide an analysis method for the activity of an automobile function, which is characterized in that the data of the whole automobile function signal is actively uploaded by an automobile CAN bus in real time, the data is received and stored in the background, and the use frequency and the use time length of the automobile function are obtained through data operation, so that the activity of the function is obtained, the real-time performance and the whole sample quantity of the data are realized, the analysis results of various dimensions CAN be formed, and the requirements of products and operators are met.

The technical scheme of the invention is as follows:

an analysis method for the activity of the functions of an automobile comprises the following steps:

step 1, original data acquisition

The original data source is from the whole bus, and the data is actively reported by the vehicle end, wherein the data comprises signal names, signal states, time stamps and the like; and synchronizing the original data into the HDFS of the private cloud at regular time, wherein the original data are in json format.

Step 2, data processing

After the raw data has been landed in the data warehouse, it is mapped into a data set (hive) table per partition per day for subsequent offline analysis.

Step 3, data analysis

Counting the times and the time length of different indexes according to service requirements, wherein each index is required to be processed by the following steps: status marking the signal of the index; merging the continuous states, and reserving the starting time and the ending time of the states; and calculating the use times and duration of the index once a day to form a result.

4. And (5) visualizing the data.

Specifically, in the step 1, the data acquisition mode is that the Tbox monitors the vehicle signal, when the state of the vehicle signal is monitored to change, the Tbox acquires the signal and stores the signal locally, and periodically uploads the signal to the cloud, that is, the original data of the vehicle function activity is synchronized to the HDFS of the private cloud at regular time, and the original data format is json.

Specifically, in the step 2, the daily data and the reissue data exist in the daily data, after the cloud acquires the data, the data is stored in each partition based on the server time, the reissue data is processed and distributed to the corresponding partition by using the sql program, and the spark computing program is activated to update the computing result of the corresponding date.

Specifically, the step 3 data analysis specifically includes:

3.1, associating each original data set (hive) table in the spark program through hiveSql, and analyzing the original data set (hive) table into a wide table; the wide table contains pieces of data in the form of a vehicle model seriescode, VIN, a timestamp, and message data candata. That is, one piece of original data is formatted into pieces of data in the form of a vehicle model seriescode, VIN, a timestamp, and message data candata by hiveSql.

3.2 reading auxiliary Table DW. Dim_active_measure data

And (3) performing state segmentation on the pieces of data in the step (3.1), wherein a field state associated index segments the state, when the measure_type is 1 or 3, the count is represented, and when the measure_type is 2, the count is represented as a duration count, and the unique measure_id can be obtained through the state and the measure_type.

3.3 Whole vehicle function Signal message grouping

After the original data is read, the original data is grouped by vin and sequenced according to the sequence of the transmission of the whole vehicle function signal message.

Further, the step 4 data visualization is to distinguish between a manual triggering function and an automatic triggering function, and for the manual triggering function, whether the function is used by a user or not is analyzed, and the low-use-rate function is analyzed from the aspects of function usability, user satisfaction and the like, so that the user satisfaction is improved; for an automatic triggering function, analyzing the use frequency of the function to obtain reliability data, and reflecting the reasonable range of the endurance test plan of the function;

the reliability data includes:

total number of functional activities: the sales and the carrying data are combined to reflect the overall situation of the function;

functionally averaged bicycle frequency: eliminating the influence of sales and carrying, and reflecting the single-car condition of the function;

the function uses the total number of vehicles: reflecting the degree of dependence of the user on the function;

trend of function usage (day, week, month): reflecting the change trend of active functions;

functional customer age group distribution: the age level ratio of the user reflecting the function;

functional usage area distribution: reflecting the geographical situation distinction of the function.

According to the method, the existing data acquisition hardware is utilized, under the condition that cost and manpower are not increased, the terminal monitors CAN bus data, uploads the whole vehicle function signal, receives and stores the data in the background, and obtains the use frequency and duration of the vehicle function through data operation, so that the function activity degree is obtained.

The invention further provides an analysis system for the activity of the functions of the automobile, which comprises the following steps:

and the acquisition module is used for: the system is used for collecting original data, wherein an original data source is from a whole bus, and data is actively reported through a 4G Tbox module, and the data comprises a signal name, a signal state, a time stamp and the like; synchronizing original data into an HDFS of a private cloud at regular time, wherein the original data are in json format;

and a data processing module: for mapping into a data set (hive) table per partition per day after the raw data lands on the data warehouse for subsequent offline analysis;

and a data analysis module: the method is used for counting the times and the time length of different whole vehicle functions according to service requirements, and each function needs to be subjected to the following processing steps: status marking the signal of the index; merging the continuous states, and reserving the starting time and the ending time of the states; calculating the use times and duration of the index once every day to form a result;

the data visualization module is used for displaying the use indexes of the functions of each vehicle in a visual form for data analysis of company products and research and development, reading the use condition of the functions and crowd distribution from each dimension, and guiding the product design and strategic direction.

Further, the data analysis module specifically includes:

a data analysis unit: associating each original data set (hive) table in the spark program through hiveSql, and analyzing the original data set (hive) table into a wide table; the wide table contains pieces of data in the form of a vehicle model seriescode, VIN, a timestamp, and message data candata.

Reading the auxiliary table data unit: and (3) carrying out state segmentation on a plurality of pieces of data of the data analysis unit, wherein a field state associated index segments a state, when the measure_type is 1 or 3, the count is 2, and a unique measure_id can be obtained through the state and the measure_type.

The whole vehicle function signal message grouping unit comprises a whole vehicle function signal message grouping unit: and grouping the read data by vin, and sequencing according to the sequence of the transmission of the whole vehicle function signal message.

The invention has the following advantages:

1. the data of the invention is actively reported by the vehicle end, namely, the data source comes from the bus of the whole vehicle and is reported in real time, the functional activity condition of the actual user can be reflected through the data, the data is processed, the analysis of various dimensions is formed, and the requirements of products and operators are met.

2. The data source comes from the whole bus, is different from the collection of the functional activity of the mobile terminal, and the data is transmitted to the whole bus, the 4G module and the background, so that the real-time performance of the data is ensured at the same time of multi-round data transmission. The data is uploaded through the whole vehicle equipment in the whole process, the uploading period is configurable, the time from the generation of the terminal equipment signal to the receiving of the server side is not longer than 1s, and the real-time performance and the integrity of the data can be ensured.

3. The method is simple, reliable and feasible, does not increase the cost additionally, collects the real behavior data of the total users carrying the Tbox by using the existing hardware equipment, and reduces the waiting period for obtaining the investigation result.

Drawings

FIG. 1 is a logic flow diagram of the present invention;

FIG. 2 is a state-labeling result in a data analysis process;

FIG. 3 is a state merging result in the data analysis process;

FIG. 4 is a table of bicycle results during data analysis;

FIG. 5 is a total number of functional activities for data visualization;

FIG. 6 is a functional average bicycle frequency for data visualization;

FIG. 7 is a functional use vehicle total for data visualization;

FIG. 8 is a functional usage trend of data visualization;

FIG. 9 is a functional customer age bracket distribution for data visualization;

FIG. 10 is a functional use area distribution for data visualization.

Detailed Description

The invention is further described below with reference to the accompanying drawings:

referring to fig. 1, the invention mainly completes the analysis of the activity of the automobile function through the cooperation of four steps of acquisition, processing, analysis and visualization:

1. raw data are collected and uploaded to hive: and monitoring the vehicle function signal through the 4G Tbox, and when the vehicle function signal meets the conditions, collecting the signal by the Tbox and storing the signal to the local, and uploading the signal to the cloud every 10 s.

2. Data processing, namely formatting the original data through hiveSql: the daily data and the reissue data exist in the daily data, after the cloud acquires the data, the data is stored in each partition according to the server time, the reissue data are processed and distributed to the corresponding partition by the sql program, and the spark computing program is activated to update the computing result of the corresponding date.

Processing of exception data:

1. terminal time abnormality (time exceeds the date of the server on the same day)

If the terminal time of a piece of data is a future time, the piece of data is not distributed or is saved in the current partition. In daily computing tasks, spark will exclude these data, screening only the data on the day for computation. By the method, the current day result can be prevented from being influenced by the terminal time abnormal data, the original data of the abnormal data can be stored, and evidence is reserved for the later abnormal data.

2. canData data repetition

This portion of dirty data exists primarily in the form that two pieces of canData exist at the same time and for the same vehicle. For this portion of data, deduplication processing is performed when spark calls hiveSql.

3. The data array object is null, and the length is 0:

aiming at the situation that the function usage data in part of the reported data is not reported, the part of the data is filtered in the spark calculation process and does not participate in the calculation of the final result.

3. Analysis: according to service requirements, counting times and time length of different whole vehicle functions, wherein each function needs to be processed as follows: status marking the signal of the index; merging the continuous states, and reserving the starting time and the ending time of the states; the number of times and the duration of the use of the function are calculated once a day, and a result is formed.

1. The ODS.log_data_uaes table raw data is read by hiveSql in the spark program and parsed into a wide table.

Specifically, one piece of original data is formatted into a plurality of pieces of data in the form of a vehicle model seriescode, VIN, a timestamp, and message data candata by hiveSql.

2. Reading the auxiliary table dw. Dim_active_measure data

The table is used for mapping the statistical result set to the time table, the field state is associated with the index segmentation state, and when the measure_type is 1 or 3, the count is represented, and when the measure_type is 2, the count is represented. The unique measure_id can be obtained by state and measure_type.

3. Message grouping

After the original data is read, the original data is grouped by vin and is ordered according to the sending sequence of the messages.

The analysis and statistics method of the third step is described by a specific index, namely the activity of the plasma generator:

the index is as follows: plasma generator status (automatic air conditioning only) ac_plasma, number of uses: the number of ac_plasmast hops from 0x0 to 0x 1; duration of time: the ac_plasmast signal goes from active (0 x 1) to inactive (0 x 0) for a single use period.

The steps are as follows:

(1) Marking status

The plasma generator state (supporting only automatic air conditioning) ac_plasma signal is 0x0, the flag state is a1, and when the signal is 0x2, the flag state is a2. When the signal jumps from 0 to 1, the signal is turned on once; when the signal jumps from 1 to 0, the signal is a use period.

Loading data after grouping sequencing, analyzing each message, firstly carrying out state division on an index AC_PlasmaSt, and setting a variable state as a1 when a field AC_PlasmaSt is= 0; if ac_plasmalst= 2, the variable state is set to a2. After parsing all messages, an array set of the index is generated. As in fig. 2.

(2) State slicing

The same continuous state is merged, and when the state state=a1 is started, starttime= 1544406765867, the data set is cyclically traversed, and when the state=a2, endtime= 15444406785867 is set. And generating a cut record: starttime=1544406676867, endtime=154444067885867, state=a1. And so on, until the length-1 record is assembled, and if the state of the length-1 record is the same as the state of the data of the last frame, and the state is used for counting times, the segment is considered invalid. If the state is used for statistical duration, the fragment is reserved.

(3) Statistics of times and duration

The number of times: state=a1, which means that the signal jumps from 0 to 1, then there are several records of state=a1 after slicing, and we consider that several records are used. Duration of time: when state=a2, the signal jumps from 1 to 0, then the endTime-startTime of each record represents the use duration of this turn-on, and the total use duration of the device today is counted by summing all the records (endTime-startTime) of state=a2 each day.

(4) Result association, generating result set

Association auxiliary table DW. Dim_active_measure attribute state, measure_type (1, 3- > represents times, 2- > represents duration) and FIG. 3 statistics table state, find attribute measure_id, generate statistics result data. Assuming that the state a1, the count-associated measure_id is 1001, the state a2, and the duration-associated measure_id is 1002, as shown in fig. 4.

4. Data visualization

For a manual triggering function, analyzing whether the function is used by a user, analyzing the low-use-rate function from the aspects of function usability, user satisfaction and the like, and improving the user satisfaction; and for the automatic triggering function, analyzing the use frequency of the function to obtain reliability data, and reflecting the reasonable range of the endurance test plan of the function.

Fig. 5-10 show various index visualizations of various functions of the vehicle, such as:

total number of functional activities, as shown in fig. 5;

functionally averaged bicycle frequency, as shown in fig. 6;

the function uses the total number of vehicles as shown in fig. 7;

trend of function usage (day, week, month) as shown in fig. 8;

functional customer age group distribution, as shown in FIG. 9;

the functions are distributed using regions, as in fig. 10.

Those of ordinary skill in the art will appreciate that: all or part of the steps of implementing the above method embodiments may be performed by program instructions and associated hardware, and the foregoing program may be stored in a computer readable storage medium.

The foregoing description of the embodiments has been provided for the purpose of illustrating the general principles of the invention, and is not meant to limit the scope of the invention, but to limit the invention to the particular embodiments, and any modifications, equivalents, improvements, etc. that fall within the spirit and principles of the invention are intended to be included within the scope of the invention.

Claims (7)

1. The method for analyzing the activity of the automobile function is characterized by comprising the following steps of:

(1) Raw data acquisition

The original data source is from the whole bus, and the data is actively reported by the vehicle end, wherein the data comprises signal names, signal states and time stamps, and the vehicle runs all the data; synchronizing original data into an HDFS of a private cloud at regular time, wherein the original data are in json format;

(2) Data processing

After the original data fall to the data warehouse, mapping the original data into a data set table according to a partition every day for subsequent offline analysis;

further comprising processing of the exception data:

processing terminal time anomaly data of which the time exceeds the date of the current day server: if the terminal time of one piece of data is the future time, the data is not distributed and is stored in the current partition, and in the daily calculation task, spark can exclude the data, and only the data on the same day is screened for calculation;

processing repeated for canData: if two pieces of canData are identical at the same time and in the same vehicle, the canData are dirty data, and duplication removal processing is carried out when spark calls hiveSql;

data with data array objects of null and length of 0 are filtered in the spark calculation process, and the data do not participate in the calculation of a final result;

(3) Data analysis

According to service requirements, counting times and time length of different whole vehicle functions, wherein each function needs to be processed as follows: carrying out state marking on the using times and duration signals of different whole vehicle functions; merging the continuous states, and reserving the starting time and the ending time of the states; calculating the use times and duration of the functions once a day to form a result;

(4) And (5) visualizing the data.

2. The method for analyzing the activity of a vehicle function according to claim 1, wherein in the step (1), the data acquisition mode is that a Tbox module monitors a vehicle signal, and when the state of the monitored vehicle signal changes, the Tbox acquires the signal and stores the signal locally and periodically uploads the signal to the cloud.

3. The method for analyzing the activity of the automotive function according to claim 2, wherein in the step (2), the daily data and the reissue data exist in the daily data, the cloud end stores the acquired data in each partition based on the server time, the reissue data are processed and distributed to the corresponding partition by the sql program, and the spark computing program is activated to update the computing result of the corresponding date.

4. A method for analyzing the activity of a vehicle function according to any one of claims 1 to 3, wherein the data analysis in the step (3) specifically includes:

(3.1) associating each original data set (hive) table in the spark program by hiveSql and resolving the original data set (hive) table into a wide table; the wide table comprises a plurality of pieces of data in the form of a vehicle model seriescode, VIN, a timestamp and message data candata;

(3.2) reading auxiliary Table data

Performing state segmentation on the pieces of data in the step (3.1), wherein a field state is associated with an index segmentation state, when the measurement_type is 1 or 3, the statistics of times is 2, the statistics of time length is 2, and a unique measurement_id can be obtained through state and measurement_type;

(3.3) Whole vehicle function Signal message grouping

After the original data is read, the original data is grouped by vin and sequenced according to the sequence of the transmission of the whole vehicle function signal message.

5. The method for analyzing the activity of a vehicle function according to claim 1, wherein the step (4) of visualizing data specifically includes:

the manual triggering function and the automatic triggering function are distinguished, whether the function is used by a user is analyzed for the manual triggering function, the low-use-rate function is analyzed from the aspects of function usability and user satisfaction, and the user satisfaction is improved; for an automatic triggering function, analyzing the use frequency of the function to obtain reliability data, and reflecting the reasonable range of the endurance test plan of the function;

the reliability data includes: total number of functional activities, average single frequency of functions, total number of functions using vehicles, functions using trend, age group distribution of functions clients, and distribution of functions using areas.

6. An analysis system for automotive functional activity, comprising:

and the acquisition module is used for: the system comprises a primary data acquisition module, a primary data source, a primary data reporting module and a primary data reporting module, wherein the primary data source is from a whole bus, and the primary data source actively reports data through the Tbox module, wherein the data comprises a signal name, a signal state and a time stamp; synchronizing original data into an HDFS of a private cloud at regular time, wherein the original data are in json format;

and a data processing module: for mapping into a data set (hive) table per partition per day after the raw data lands on the data warehouse for subsequent offline analysis; also for processing exception data, comprising: processing terminal time anomaly data of which the time exceeds the date of the current day server: if the terminal time of one piece of data is the future time, the data is not distributed and is stored in the current partition, and in the daily calculation task, spark can exclude the data, and only the data on the same day is screened for calculation; processing repeated for canData: if two pieces of canData are identical at the same time and in the same vehicle, the canData are dirty data, and duplication removal processing is carried out when spark calls hiveSql; data with data array objects of null and length of 0 are filtered in the spark calculation process, and the data do not participate in the calculation of a final result;

and a data analysis module: the method is used for counting the times and the time length of different whole vehicle functions according to service requirements, and each function needs to be subjected to the following processing steps: carrying out state marking on the using times and duration signals of different whole vehicle functions; merging the continuous states, and reserving the starting time and the ending time of the states; calculating the use times and duration of the index once every day to form a result;

and the data visualization module is used for displaying the use index of each vehicle function in a visual form.

7. The system for analyzing the activity of a vehicle function according to claim 6, comprising: the data analysis module specifically comprises:

a data analysis unit: associating each original data set (hive) table in the spark program through hiveSql, and analyzing the original data set (hive) table into a wide table; the wide table comprises a plurality of pieces of data in the form of a vehicle model seriescode, VIN, a timestamp and message data candata;

reading the auxiliary table data unit: the method comprises the steps of performing state segmentation on a plurality of pieces of data of a data analysis unit, segmenting a field state associated index, counting the number of times when the measure_type is 1 or 3, counting the duration when the measure_type is 2, and obtaining a unique measure_id through the state and the measure_type;

the whole vehicle function signal message grouping unit comprises a whole vehicle function signal message grouping unit: and grouping the read data by vin, and sequencing according to the sequence of the transmission of the whole vehicle function signal message.

Images